Models, providers and routing

Hosted and self-hosted providers

Cloud APIs and your own GPU servers behind one endpoint: OpenAI, Anthropic, Bedrock, Vertex, Gemini, vLLM, Ollama and more.

What it is

Native adapters for OpenAI-compatible APIs, Anthropic, AWS Bedrock, Google Vertex AI, Google Gemini, Ollama, vLLM, llama.cpp and Hugging Face TEI. Any other OpenAI-compatible server, such as LM Studio, works through the generic adapter.

Why you want it

Most organizations use a mix of cloud APIs and their own GPU servers. One gateway in front of all of them gives one set of controls and one bill view.

One OpenAI client with one Janus key calls claude-sonnet, which goes to a cloud provider, and llama-local and bge, which run on your own GPUs. client = OpenAI( base_url= "https://ai.corp/v1", api_key="je_7f3c…") client.chat( model="claude-sonnet") client.chat( model="llama-local") client.embed(model="bge") Janus CloudOpenAIAnthropicBedrock Your GPUsOllamavLLMTEI One key, one endpoint, every model. The model name decides where each request goes.

How it works

  • Registered adapter types: openai_compatible, anthropic, bedrock, vertex, gemini, ollama, vlm, llama_cpp, tei (plus codex and copilot used by personal subscriptions)
  • Upstream credentials encrypted at rest with AES-256-GCM
  • Bedrock and Vertex use your own cloud account and credentials
  • Google Gemini uses an AI Studio key, meters thinking tokens as output and records implicit cache hits

In the product

Janus Upstreams screen with demo data
Upstreams

Each provider connection with its adapter, endpoint, health and model count.

Actual Janus interface · Demo identities and synthetic usage

See it on your own network.

The Community edition is free for up to 25 people. The 30-day Business trial unlocks every Business feature.